Network metrics to localize the Seizure Onset Zone: A Directed Functional Connectivity Study Using MEG Data (P2-1.004)

Abstract

<h3>Objective:</h3> We aimed to analyze ictal networks via magnetoencephalography-based (MEG) directed functional connectivity (FC), to better guide intracranial EEG (iEEG) studies. <h3>Background:</h3> Epilepsy is a network disorder. Thus, the importance of integrating connectivity-based approaches during epilepsy presurgical evaluation. Studies applying MEG non-directed FC evidenced a good correlation with post-surgical outcomes. Directed FC, such as Spectral Bivariate Granger Causality (SBGC), has the advantage of differentiating probable sinks from hubs in networks. <h3>Design/Methods:</h3> Ictal MEG and MRI data from eight drug-resistant epilepsy patients (four pediatric, six females) were included. Source space SBGC was performed and grouped in frequency bands. Nodes with higher influence in outgoing signals (hubs) and receiving signals (authority) were identified. The top 2 hub and authority nodes were selected and compared to clinical data. <h3>Results:</h3> Seven patients had iEEG, three of which had subsequent epilepsy surgery. Another patient underwent direct surgery. Two patients continued presenting seizures after resection. All others were seizure-free after 1-month and 18-month follow-ups. For all patients, the irritative zone (80%) or nearby regions (20%) were captured by one of the top two hub nodes in the highest power frequency band and the delta band. For the patients with seizure freedom, at least one of the top two hub nodes included the Seizure Onset Zone (SOZ). In the patient with an 18-month Engel 1A classification, both top hub nodes were included in the resection area. For an adult patient with multifocal SOZs, the highest authority node represented another less active SOZ. <h3>Conclusions:</h3> The regions identified by our connectivity metrics demonstrated overlap with regions identified with iEEG in the majority of the patients, and those patients in which resection included a hub node had favorable post-surgical outcomes. We contend that network metrics using SBGC and MEG data can be useful to better understand epilepsy dynamics and enhance epilepsy management in clinical practice. <b>Disclosure:</b> Dr. Cardoso Da Fonseca has nothing to disclose. Ms. Askari has nothing to disclose. Dr. Proskovec has nothing to disclose. Dr. Alick-Lindstrom has nothing to disclose. Dr. Lowden has nothing to disclose. Dr. Talai has nothing to disclose. The institution of Joseph Maldjian has received research support from NIH. Dr. Davenport has nothing to disclose.